On 12.01.2012 21:13, Philip Martin wrote:
> Stefan Küng<tortoisesvn_at_gmail.com> writes:
>> Question is: do we need the notifications for deleting empty dirs in a
>> dry-run? If yes, then this get's complicated:
>> the error is thrown from svn_wc_walk_status() called in
>> check_dir_empty. I could just not call that function in case of a
>> dry-run, or catch that specific error and go on. But then the
>> notifications wouldn't be completely accurate anymore.
>> So assuming we don't want to extend svn_wc_walk_status to take a
>> dry-run param, the notifications for deleting empty dirs wouldn't be
>> Should I just change delete_empty_dirs() to not use the dry_run param
>> and not get the notifications for those in a dry-run?
>> I think no notifications is better than inaccurate ones (at least in a
> I'm not clear what problem the patch is triggering. I assume the patch
> deletes one or more files but then what happens?
> At present a --dry-run patch that deletes a file will track the
> deletions and notify about a directory that would be empty and deleted,
> we don't want to lose that feature.
I'm not sure how this is triggered exactly. If you check the mailing
list thread on the TSVN mailing list:
you can see that both the working copy and the patch to reproduce the
issue are quite big.
From what I could find out I think the problem is when the patch
removes all versioned files in a folder, but adds new files in that same
In a real run, the file is added to the working copy and the folder is
not empty anymore. In a dry-run however, the folder is marked as
removed, but then the still unversioned file is still there and that's
when the error
The Node ..../newDirectory was not found
download the file
(save as test.patch)
$ svn co https://triplea.svn.sourceforge.net/svnroot/triplea/trunk wc -r3375
$ svn patch test.patch wc --dry-run
oo // \\ "De Chelonian Mobile"
(_,\/ \_/ \ TortoiseSVN
\ \_/_\_/> The coolest Interface to (Sub)Version Control
/_/ \_\ http://tortoisesvn.net
Received on 2012-01-12 21:49:33 CET